The managed run
The run, not the bytes. The engine is free; operating it well - at scale, with proof, for auditors - is what you pay for.
Hosted execution
Your migrated jobs run on the same engine that validated them, on provisioned CPU and GPU capacity, with the seasonal pre-warm and capacity plan rehearsed before peak days.
The dashboards
Every run lands in your portal: job, tier, scale, wall time, and the gate's verdict - IDENTICAL, MISMATCH with the diff, or UNVALIDATED, which is never proven. Run records are checksummed; a tampered record is flagged, not silently accepted.
The audit trail
The dual-run reconciliation stream during shadow running, the per-run evidence archive, and the licence line that names the licensee on every keyed run. The artefact your steering committee and your auditor actually want.
Pricing
By job tier and SLO tier, billed monthly by Stripe. Indicative bands: 10-50k per month by SLO tier, scaled by estate size; a representative 100M-transaction nightly batch prices in the low hundreds of dollars per month on commodity hardware. Exact numbers come from profiling your estate - which is what the proof trial measures.
Why the managed run exists
The open-source model works when operating the software is the real burden. A byte-exact migration runtime is exactly that: capacity, pre-warm, dual-run reconciliation, audit evidence, SLOs. We built the engine; we are best placed to run it, and the subscription funds the open standard you rely on.